The present invention relates to the detection of an endpoint in the etching of a substrate.
In the processing of a substrate to fabricate electronic devices, such as integrated circuits and displays, etching processes are carried out to etch materials on the substrate to form patterned features that form components of the electronic devices. For example, the patterned features may comprise gates, vias, contact holes, or interconnect lines. Typically, a patterned mask of etch-resistant features comprising resist or hard-mask materials is formed on the substrate, and exposed areas of the substrate between the etch-resistant features are etched to form the patterned features.
During the etching process, an endpoint determination method is used to evaluate and control etching progress through the substrate, such as to stop or change etching parameters at a predetermined etch depth. In interferometric endpoint determination methods, as illustrated in FIG. 1 (prior art), light beams 76a,b are directed onto the substrate 10 and the beams are reflected from the substrate 10 to form reflected light beams 78a,b. Constructive and destructive interference of the reflected light beams 78a,b modulate the total or summed reflected light 78 over time to generate interference fringes having intensity maxima and minima. The reflected light 78 is detected by a detector that generates a reflection signal, and the reflection signal is monitored to determine an endpoint of the etching process. The reflection signal exhibits maxima and minima peaks that correspond to interference fringes that arise from primary reflections 78a from the surface of the etch layer 22 and secondary surface reflections 78b from the surface of the mask features 24, as well as from other reflections from internal interfaces between layers of the substrate 10. By counting these interference fringes, it can be determined when an etching process endpoint has been reached, such as a particular etch depth or etch rate, that occurs after a predetermined number of fringes are counted.
However, the effective signal-to-noise ratio of the interference fringes of the reflected light 78 is relatively low because the intensity of the interference signal is also affected by the internal reflections that arise from interfaces within the substrate 10. For example, a portion 76b of the light beam 76 that is incident on the mask features 24 is also partially transmitted to the interface 26 between the mask features 24 and the etch layer 22. The reflection 78c of the light beam 78 from the interface beneath the mask features 24 undesirably interferes with the surface reflections 78a,b to reduce the overall strength or intensity of the reflected light beam 78. This reduction of the reflected light beam intensity hinders endpoint detection by decreasing the effective signal-to-noise ratio of the interference fringes.
As semiconductor devices are fabricated to have increasingly smaller dimensions, it is desirable to detect endpoint with better precision, to stop or change processing when the desired small dimension is reached. However, interface reflections 78c from below the mask features 24 effectively limit the precision and accuracy of endpoint detection by adding noise to the reflection signal. While this noise can be partially removed using filters such as bandpass filters, the filters increase the complexity of the endpoint detection system and often do not reduce the noise to a sufficiently low level. Noise levels are particularly deleterious when etching devices having shallower or more precise depths, or when etching a thin layer on the substrate and stop etching in due time without etching through the thin layer. Accordingly, it is desirable to interferometrically determine the endpoint during substrate processing with higher precision and better signal to noise ratio